EVALUATION OF EDGE-PRESERVING SMOOTHING FILTERS FOR DIGITAL IMAGE MAPPING

The digital orthophoto has the potential for becoming a standard sourc e of high quality imagery for cartographic applications. With rectifie d, high resolution imagery available in digital form, a variety of ima ge processing techniques can be applied to aid extraction of cartograp hic features. This paper describes research on one such type of proces sing, specifically a performance evaluation of six edge-preserving noi se smoothing filters. This evaluation is the first to be performed in the context of cartographic feature extraction from digital imagery an d to include a summary figure-of-merit calculated from measurements on the filtered digital imagery. The results suggest that the best compr omise between smoothing and edge preservation is obtained with the com bination of the sigma filter followed by the contrast-enhancing biased sigma filter, although this approach is computationally more expensiv e than most of the other filters.